RTDYLDMemoryManager.

llvm-svn: 192516

attribute.  Patch by chris.wailes@gmail.com; reviewed and edited by delesley.

llvm-svn: 192515

This fixes a problem from a previous check-in where a return value was omitted.

Previously the remote/stubs-remote.ll and remote/stubs-sm-pic.ll tests were reporting passes, but they should have been failing.  Those tests attempt to link against an external symbol and remote symbol resolution is not supported.  The old RemoteMemoryManager implementation resulted in local symbols being used for resolution and the child process crashed but the test didn't notice.  With this check-in remote symbol resolution fails, and so the test (correctly) fails.

llvm-svn: 192514

tests_typestate attribute.  Patch by chris.wailes@gmail.com.

llvm-svn: 192513

function parameter that has array type.  Such a parameter will be treated as
a pointer type instead, resulting in a missing begin function error is a
suggestion to dereference the pointer.  This provides a different,
more informative diagnostic as well as point to the parameter declaration.

llvm-svn: 192512

gdb-format a (as in p/a) would fail as it needed to set a byte size (unsurprisingly enough)
This should be acknowledged by the condition check and not cause a failure

llvm-svn: 192511

Fixed Module::ResolveSymbolContextForAddress() to be able to also look in the SymbolVendor's SymbolFile's ObjectFile for a more meaningful symbol when a symbol lookup finds a synthetic symbol from the main object file. This will help lookups on MacOSX as the main executable might be stripped, but the dSYM file always has a full symbol table.

llvm-svn: 192510

llvm-svn: 192509

This allows the callable_when attribute to be attached to destructors.
Original patch by chris.wailes@gmail.com, reviewed and edited by delesley.

llvm-svn: 192508

Summary:
This way we avoid breaking code which uses unknown preprocessor
directives with long string literals. The specific use case in
http://llvm.org/PR17035 isn't very common, but it seems to be a good idea to
avoid this kind of problem anyway.

Reviewers: djasper

Reviewed By: djasper

CC: cfe-commits, klimek

Differential Revision: http://llvm-reviews.chandlerc.com/D1813

llvm-svn: 192507

NS_ENUM/NS_OPTIONS macros when typedef declaration
precedes enum declaration. // rdar://15200915 

llvm-svn: 192506

Changed clang-format-diff.py to output diff by default. Added -i option to apply changes to files instead.

Summary:
"svn diff|clang-format-diff.py" will just output the diff.
Now it's possible to use:
  svn diff|clang-format-diff.py|patch -p0
as an equivalent to:
  svn diff|clang-format-diff.py -i
;)

Reviewers: djasper

Reviewed By: djasper

CC: cfe-commits

Differential Revision: http://llvm-reviews.chandlerc.com/D1840

llvm-svn: 192505

llvm-svn: 192504

We now correctly format:
  void SomeFunction(int param1,
  #ifdef X
                    NoTemplate param2,
  #else
                    template <
  #ifdef A
                        MyType<Some> >
  #else
                        Type1, Type2>
  #endif
                    param2,
  #endif
                    param3) {
    f();
  }

llvm-svn: 192503

We're not ready to use it everywhere we use a win32 triple yet.

llvm-svn: 192502

llvm-svn: 192501

llvm-svn: 192500

llvm-svn: 192499

Use -no-struct-path-tbaa to turn it off.
This is the same as r191695, which was reverted because it depends on a
commit that has issues.

llvm-svn: 192497

Calling convention attributes can add sugar to methods that we have to
look through.  This fixes an assertion failure in the provided test
case.

llvm-svn: 192496

llvm-svn: 192495

llvm-svn: 192494

[libclang] Introduce clang_Type_getCXXRefQualifier whichexposes ref-qualifier information of function type.

Patch by Che-Liang Chiou!

llvm-svn: 192493

llvm-svn: 192492

Added a way to set hardware breakpoints from the "breakpoint set" command with the new "--hardware" option. Hardware breakpoints are not a request, they currently are a requirement. So when breakpoints are specified as hardware breakpoints, they might fail to be set when they are able to be resolved and should be used sparingly. This is currently hooked up for GDB remote debugging. 

Linux and FreeBSD should quickly enable this feature if possible, or return an error for any breakpoints that are hardware breakpoint sites in the "virtual Error Process::EnableBreakpointSite (BreakpointSite *bp_site);" function.

llvm-svn: 192491

In certain macros or incorrect string literals, the token stream can
contain 'unknown' tokens, e.g. a single backslash or a set of empty
ticks. clang-format simply treated them as whitespace and removed them
prior to this patch.

This fixes llvm.org/PR17215

llvm-svn: 192490

<rdar://problem/15188389>

llvm-svn: 192489

llvm-svn: 192488

llvm-svn: 192487

llvm-svn: 192486

llvm-svn: 192484

This doesn't change the meaning of the output, but makes look right. PR17539.

llvm-svn: 192483

When if converting something like:
true:
   ... = R0<kill>

false:
   ... = R0<kill>

then the instructions of the true block must not have a <kill> flag
anymore, as the instruction of the false block follow and do still read
the R0 value.
Specifically this patch determines the set of register live-in in the
false block (possibly after simulating the liveness changes of the
duplicated instructions). Each of these live-in registers mustn't be
killed.

llvm-svn: 192482

Contains a set of live register (units) and code to move forward and
backward in the schedule while updating the live set.

llvm-svn: 192481

[DAGCombiner] Load slicing test case: attempt to really fix the buildbots (used sse4.2 instead of avx!).

<rdar://problem/14477220>

llvm-svn: 192480

llvm-svn: 192479

llvm-svn: 192478

Really fix CHECK-LABEL and CHECK-DAG interaction. This actually just restores the initial implementation that was in r186162 but got lost in some subsequent refactoring. More explicit variable names and comments are present now to hopefully prevent repeat regression, as well as another test.

llvm-svn: 192477

This should fix the buildbots.

Original commit message:
[DAGCombiner] Slice a big load in two loads when the element are next to each
other in memory and the target has paired load and performs post-isel loads
combining.

E.g., this optimization will transform something like this:
a = load i64* addr
b = trunc i64 a to i32
c = lshr i64 a, 32
d = trunc i64 c to i32

into:
b = load i32* addr1
d = load i32* addr2
Where addr1 = addr2 +/- sizeof(i32), if the target supports paired load and
performs post-isel loads combining.

One should overload TargetLowering::hasPairedLoad to provide this information.
The default is false.

<rdar://problem/14477220>

llvm-svn: 192476

llvm-svn: 192474